Contents 38I.38II.Approaches for reconstructing refugia: strengths, limitations and recent advances39III.46IV.47V.48VI.4949References49 Summary Climate refugia, locations where taxa survive periods of regionally adverse climate, are thought to be critical for maintaining biodiversity through the glacial–interglacial climate changes of the Quaternary. A critical research need is to better integrate and reconcile the three major lines of evidence used to infer the existence of past refugia – fossil records, species distribution models and phylogeographic surveys – in order to characterize the complex spatiotemporal trajectories of species and populations in and out of refugia. Here we review the complementary strengths, limitations and new advances for these three approaches. We provide case studies to illustrate their combined application, and point the way towards new opportunities for synthesizing these disparate lines of evidence. Case studies with European beech, Qinghai spruce and Douglas‐fir illustrate how the combination of these three approaches successfully resolves complex species histories not attainable from any one approach. Promising new statistical techniques can capitalize on the strengths of each method and provide a robust quantitative reconstruction of species history. Studying past refugia can help identify contemporary refugia and clarify their conservation significance, in particular by elucidating the fine‐scale processes and the particular geographic locations that buffer species against rapidly changing climate.

The paucity of low- and middle-elevation paleoecologic records in the Northern Rocky Mountains limits our ability to assess current environmental change in light of past conditions. A 10,500-yr-long vegetation, fire and climate history from Lower Decker Lake in the Sawtooth Range provides information from a new region. Initial forests dominated by pine and Douglas-fir were replaced by open Douglas-fir forest at 8420 cal yr BP, marking the onset of warmer conditions than present. Presence of closed Douglas-fir forest between 6000 and 2650 cal yr BP suggests heightened summer drought in the middle Holocene. Closed lodgepole pine forest developed at 2650 cal yr BP and fires became more frequent after 1450 cal yr BP. This shift from Douglas-fir to lodgepole pine forest was probably facilitated by a combination of cooler summers, cold winters, and more severe fires than before. Five drought episodes, including those at 8200 cal yr BP and during the Medieval Climate Anomaly, were registered by brief intervals of lodgepole pine decline, an increase in fire activity, and mistletoe infestation. The importance of a Holocene perspective when assessing the historical range of variability is illustrated by the striking difference between the modern forest and that which existed 3000 yr ago.

Abstract1. In northern Idaho (USA), more than 100 vascular plant species are disjunct >160 km from their main distribution along the Pacific Northwest coast. It remains unclear whether most species within this interior forest disjunction, including Tsuga mertensiana, survived the last glacial period in a north-Idaho refugium or whether these species colonized the region via long-distance dispersal during the Holocene.2. Sediment cores were extracted from three mid-to high-elevation lakes within T. mertensiana-dominated forests in the Northern Rocky Mountains of Idaho.Pollen and macrofossils were used to reconstruct forest composition, determine the timing of T. mertensiana establishment, examine the hypothesis that the region was a glacial refugium, and contrast how climate, competition and/or dispersal limitation have influenced its modern distribution.3. The modern distribution of T. mertensiana was analysed by constructing a range map and modelling the potential species distribution. The presence of outlier populations surrounding the Idaho disjunction along with broad areas of unoccupied suitable habitat indicates that the range of T. mertensiana is currently expanding. To assess the accuracy of T. mertensiana pollen at detecting its range limit, a network of pollen surface samples was used to analyse the probability of detecting T. mertensiana pollen as a function of distance from its geographical range limit. Consistent T. mertensiana pollen occurrence at ≥1% abundance is likely only within 42 km of its range limit. 4.Tsuga mertensiana first appears in the pollen and macrofossil record at the highestelevation site at c. 4,100 cal year bp, then at the next highest-elevation site at c. 1,600 cal year bp, and last at the mid-elevation site at 800 cal year bp. Tsuga mertensiana pollen occurs continuously at ≥1% at all three sites by c. 300 cal year bp suggesting regional presence. The timing of arrival suggests that T. mertensiana is a recent component of the forests of Idaho, having arrived during the Holocene via long-distance dispersal from coastal populations over 160 km away. Synthesis.Comparison with palaeoclimate reconstructions from the broader region suggests that climate was a greater limiting factor than dispersal in the Holocene establishment in the interior, indicating little difficulty overcoming a large dispersal barrier. However, T. mertensiana remained at low abundances for millennia untilPaper previously published as Standard Paper

Aim: Understanding how climate refugia and migration over great distances have facilitated species survival during past climate changes is crucial for evaluating contemporary threats to biodiversity, particularly in the face of dispersal barriers. We address this longstanding question on the refugial origins and post-glacial development of mesic forests.Location: Pacific Northwest, North America.Taxon: Mountain hemlock (Tsuga mertensiana) and western redcedar (Thuja plicata).Methods: Range-wide genotyping-by-sequencing (ddRADseq) of both study species and a pollen reconstruction of mountain hemlock presence over the last 20,000 years.Results: Mountain hemlock occurred in two coastal populations (Oregon and Washington) during the glacial maximum, each of which dispersed to the interior (Idaho and British Columbia) during the Holocene. These populations spread in the direction of dominant winds across a barrier of dry, rain-shadowed valleys. In contrast, for western redcedar, we infer four disparate refugia during the glacial maximum: southern (California), central (Washington), interior (Idaho), and northern (Haida Gwaii islands). Main conclusions:Despite the presence of pre-dispersed refugial populations, the majority of the redcedar distribution was colonized by the central population. The history for these two key conifers contrast with many recent studies emphasizing the role of cryptic refugia in colonizing modern species ranges.

Larix laricina (eastern larch, tamarack) is a transcontinental North American conifer with a prominent disjunction in the Yukon isolating the Alaskan distribution from the rest of its range. We investigate whether in situ persistence during the last glacial maximum (LGM) or long‐distance postglacial migration from south of the ice sheets resulted in the modern‐day Alaskan distribution. We analyzed variation in three chloroplast DNA regions of 840 trees from a total of 69 populations (24 new sampling sites situated on both sides of the Yukon range disjunction pooled with 45 populations from a published source) and conducted ensemble species distribution modeling (SDM) throughout Canada and United States to hindcast the potential range of L. laricina during the LGM. We uncovered the genetic signature of a long‐term isolation of larch populations in Alaska, identifying three endemic chlorotypes and low levels of genetic diversity. Range‐wide analysis across North America revealed the presence of a distinct Alaskan lineage. Postglacial gene flow across the Yukon divide was unidirectional, from Alaska toward previously glaciated Canadian regions, and with no evidence of immigration into Alaska. Hindcast SDM indicates one of the broadest areas of past climate suitability for L. laricina existed in central Alaska, suggesting possible in situ persistence of larch in Alaska during the LGM. Our results provide the first unambiguous evidence for the long‐term isolation of L. laricina in Alaska that extends beyond the last glacial period and into the present interglacial period. The lack of gene flow into Alaska along with the overall probability of larch occurrence in Alaska being currently lower than during the LGM suggests that modern‐day Alaskan larch populations are isolated climate relicts of broader glacial distributions, and so are particularly vulnerable to current warming trends.